SUPERCOMPUTER TECHNOLOGIES IN MATHEMATICAL...

SUPERCOMPUTER TECHNOLOGIES IN

MATHEMATICAL MODELLING (SCTEMM-2013)

Yakutsk, Russia July 8-11, 2013

Author: Dr. Atul M. GonsaiBhargavi Goswami,

Kar Uditnarayan

Dept. of Computer Science, Saurashtra University, Rajkot-360005, Gujarat, India.

Jul 7, 2013 Saurashtra University Rajkot 1

Study of Supercomputer’s Architecture, Application and Its Future UseStudy of Supercomputer’s Architecture, Application and Its Future Use

1. Introduction2. History of Super-Computer :

PARAM Yuva II3. Requirement of Supercomputer4. Configuration of Cray Supercomputer5. Application of Supercomputer6. Future of Supercomputer7. Conclusion8. References



1. Introduction2. History of Super-Computer :: PARAM

Yuva II3. Requirement of Supercomputer4. Configuration of Cray Supercomputer5. Application of Supercomputer6. Future of Supercomputer7. Conclusion8. References



• Supercomputer is a computer with high performance than personal computers.

• It has something superior than normal computer• Used for scientific research such as NASA and industrial

fields such as in IBM.• Provides high speed calculation for large amount of data.• Supercomputers were developed in the 1960s at Control

Data Corporation (CDC) by Seymour Cray and later on known as Cray Research.

• Initially less but at the end of the 20th century it started using thousands of processors along with parallel supercomputers.

• Figure-1 shows the Blue Gene/P supercomputer that runs over 250000 processors using normal data centre air-condition.

• It is grouped in 72 racks that are connection through high-speed network.



IntroductionIntroduction


Study of Supercomputer’s Architecture, Application and Its Future UseStudy of Supercomputer’s Architecture, Application and Its Future UseCDC 6600

History of Super-Computer : History of Super-Computer : PARAM Yuva IIPARAM Yuva II• Seymour Cray founded Control Data

Corporation (CDC).• The first supercomputer was CDC 6600,

developed in 1964 using circuits and transistors.

• Cray founded its own company in 1970s.• In 1980, use of processors in supercomputer

was limited, by 1990, processors used were in thousands in Japan and USA.

• It progressed in first decade of 21st century by using 60,000 processors in supercomputer that reached to petaflop performance levels.



History of Super-Computer : History of Super-Computer : PARAM Yuva IIPARAM Yuva II

• Evolution of supercomputers includes 4 stages: (Name : Era : Performance)– Budded stage (1960-1975), Mflop/s (106)– Vector stage (1976-1989), Gflop/s (109)– MPP stage (1990-2000), Tflop/s (1012)– Massive Parallel Processor (MPP) & Cluster

stage (2000-2010), Pflop/s (1015).• Where, flop/s means Floating Point

Operations / Second.



History of Super-Computer : History of Super-Computer : PARAM Yuva IIPARAM Yuva II• In India:

– PARAM 8000 was the first supercomputer developed by “Centre for Development of Advanced Computing” (C-DAC) at Pune at 1990s and installed in 1991.

– After that the improved PARAM series was developed one by one which are below.

• PARAM 8600• PARAM 9900/SS• PARAM 10000• PARAM Padma• PARAM Yuva• PARAM Yuva II

– Let us talk about the last one, PARAM Yuva II



History of Super-Computer : History of Super-Computer : PARAM Yuva IIPARAM Yuva II

• PARAM Yuva II was developed in C-DAC in a period of 3 months at a cost of 16 Crore.

• The PARAM Yuva II released with 524 Tflop/s.• 10 times faster than the PARAM Yuva.• It uses 35% energy less than the PARAM Yuva.• It ranked at 33rd in the November 2012 list of Top Green 500

Supercomputers of the world in terms of energy efficiency.• Used for research in space, bioinformatics, weather forecasting,

seismic data analysis, aeronautical engineering, scientific data processing and pharmaceutical development

• Uses Infiband N/w for efficient data transfer.• The storage capacity of this system is 200 tera bytes. • It also provides a tool for complex programs. • The PARAM Yuva II stands on 62nd position in world’s Top 500

supercomputers



Requirement of SupercomputerRequirement of Supercomputer• Solving high level of complex problems.• Weather forecasting for global climate changes, • Quantum mechanics to study physical systems at

atomic level.• Also uses are in military, chemical composition

and polymer research, aerospace and cosmology, energy security, healthcare and medicine, manufacturing, environment protection, etc.

• Have been used to determine the impact they would have in particular region by simulation nuclear bomb detonation.



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerWe would discuss:1. Architecture2. Input / Output3. Computation Section4. Memory5. Operating register6. Hardware Design



Fig. Cray – 1



Configuration of Cray SupercomputerConfiguration of Cray Supercomputer• Cray-1 supercomputer consist of CPU, one or more minicomputer

console, power and cooling equipments and mass storage disk subsystem.

• CPU holds the I/O section, computation and memory of the computer.

• A minicomputer uses for either as a job entry station or maintenance control unit.

• 1960s, we used parallel computing to achieve superior computational peak load but later on, it started using massively parallel system.

• Grid system is an example of massively parallel system.• Cray-1 supercomputer was developed in year 1976 (previous fig). • It was first general purpose supercomputer which used Vector

Processor machines from Cray Research Inc.• Cray-1 was the fastest supercomputer whose computational rate

was 138 million floating-point operations per second.• Then came, Cray-XMP and Cray-2 which were used to solve large

scale problems in science and engineering.



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerArchitecture:• Cray-1 supercomputer uses register oriented

vector processor.• vector registers are used as data cache between

memory and functional unit (ALU).• Vector registers are prepared to hold up to 64 bit

vector elements• The performance depends on how fast the data

transfer occurs through registers.• Division of architecture:

a) Moduleb) Coolingc) Maintenance



Configuration of Cray SupercomputerConfiguration of Cray Supercomputera) Module:

– Cray-1 uses modules that were laid horizontally and stacked in tower formation as shown in previous figure.

– Each module has a 5 layered circuit board.– These 5 layers are divided into two parts, first part has

2 outer layers and second part has 3 inner layers.– It used up to 288 IC’s and so generated enormous

heat.– Has problem of signal transmission: Cray-1 passes

the standing waves and that’s why, it could not reach up to the top of the level and thus reflected with weak signals.



Configuration of Cray SupercomputerConfiguration of Cray Supercomputera) Cooling:

– new cooling technology was developed.– The vertical aluminium/stainless steel bars lined each

column wall in each chassis. – This technology was Freon based technology and it

had 40 tone cooling capacity. – It is flowing from top to bottom in each of the tower.

a) Maintenance:– Initially, the cray-1 supercomputer Mean Time

between Failure (MTBF) was order of 50 hours. – It requires 2 hour maintenance daily and the

additional time was devoted in the weekend.– This gives better performance even while working for

full day.



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerInput / OutputInput / Output:• Cray-1 supercomputer has 24 channels for

input and output. • From that 12 are used for input and 12 for

output. • For 16-bit channel, data transfer rates of 160

million bits per second can be achieved, and the higher rates are dependent on peripheral devices.

• I/O and CPU share a memory using single port



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerComputation SectionComputation Section:• Two sections: 1) Functional Unit & 2) Data Structure1. Functional Unit• The cray-1 supercomputer used 12 functional units (CPU) at a time

which allowed parallel processing. • It means that the CPU fetches the next data as well as run the

current task.• Because of CPU allowed parallel processing, it increased process

speed and CPU dead time was minimized.• These 12 functional units were divided into 4 groups those are

Address unit, Scalar Unit, Vector Unit and Floating-point Unit.• The first three groups are used to conjunct with one of the primary

register type to support modes of processing of address, scalar and vector unit.

• The fourth group is used to support either scalar or vector operation and it will accept the operand from it then deliver it to scalar or vector accordingly.

• See next table.



No. Of Functional Unit

Use For

3 Integer Addition1 Integer Multiply2 Shift2 Logical1 Floating Addition1 Floating Multiply1 Reciprocal approx.1 Population Count

2. Data Structure•The cray-1 use ASCII character representation with 64-bit word means 8 characters at a time. •The numeric representation is either in 64-bit floating-point form using sign bit and biased exponent or in 2’s complements form (24-bit or 64-bit).



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerMemoryMemory:• Cray-1 memory was organized in 16 banks; each bank had 72

modules.• Up to 5,24,288 words can be arranged in 8 banks and a word

consists of 64 data bits & check bits.• The bank cycle time is 50 nano second. • It is the time required to remove or insert an element of data in

memory.Operating register:Operating register:• The cray-1 has 5 registers. • From that 3 are primary registers and 2 are intermediate registers. • Primary registers are used by functional unit. • Intermediate registers are not accessible by functional unit but are

used as buffer in between functional unit and memory. • In addition, there are supporting register available in CPU. • Channel address register and channel limit register for each I/O

Channel



Configuration of Cray SupercomputerConfiguration of Cray SupercomputerHardware Design:• It was hand-wired and its circuits were

arranged in a three-quarters circle.• As a result, it helped to increase the speed at

which signals travel from one part to another part of the computer.

• Hefty power supplies are located below each of the circuit bays and padding them suggested a piece of lounge furniture.



Configuration of Cray SupercomputerConfiguration of Cray Supercomputer

So, we have discussed:1. Architecture2. Input / Output3. Computation Section4. Memory5. Operating register6. Hardware Design



Application Of SupercomputerApplication Of Supercomputer• Currently supercomputer applications are being developed in two

directions: – Capability Computing and – Capacity Computing

• Capability computing:– target is to focus on the critical scientific challenges to reduce cost and time for

execution• Capacity computing

– focuses on communication intensive applications, which means that finishing as many tasks as possible in the specified time domain

• ANSYS application is an example of supercomputer application which uses Cray cX1 supercomputer.

• Cray cX1 system is mainly a cluster having 8 blades and that is in dedicated networks.

• Each blade has two Intel processors. That configured with either Windows HPC Server 2008 or Linux Operating System.

• One blade is designed as a "head node" or "login node" which has different properties then remaining blades and that remaining blades are called as "compute nodes"



Future Of SupercomputerFuture Of Supercomputer• Exascale (1018 FLOPs) supercomputer will be available in 2018.• The Japanese HPC community has identified the following 6 issues:

– Improving Energy-efficiency– Using memory hierarchy to reduce the traffic of off-chip memory– Exploiting parallelism of millions of processing cores– Heterogeneous computing for accelerating the performance– Dependability of the millions of processor cores and their interconnections– Productivity for the complicated system

• The general purpose supercomputing has to face key challenges that are given as follows– Maintaining good vector/scalar performance balance– Supporting scalability– Increasing Memory capacity and performance– Providing high performance I/O and networking

• Energy-efficiency is the most difficult challenge for future supercomputing.• Current supercomputer energy consumption is equal to supply power to

5000 homes.• The future supercomputer must consume less amount of energy, as well

as generate less amount of heat.



ConclusionConclusion• There is only one mother supercomputer and

under that there are numbers of sub supercomputer.

• Supercomputers use massively parallel computation which makes its task execution faster.

• Hardware parts are also big in size so, it requires large accommodation area.

• Supercomputers use huge number of ICs so it generates more heat and it has to be maintained by using some cooling system.

• Supercomputers require daily maintenance.




References1. Xianghui XIE, Xing FANG and Sutai HU, Dong WU, Evolution of supercomputers, Jiangnan Institute of

Computing Technology, Wuxi 214083, China, page no 428-436. G. Bianchi, “Performance analysis of the IEEE 802.11 Distributed Coordination Function”, IEEE Journal on selected areas in communications, Vol. 18, No. 3, March 2000, pp. 535 - 547.

2. Book by Hoffman, Allan R, “Supercomputers: direction in technology and applications”, National Academics, page no 21, 29.

3. Book by John A. N. Lee, John Impagliazzo, History of computing in education, Kluwer Academic Publishers, WCC Toulouse, 18th IFIP World Computer Congress, page no 172.

4. Sao-Jie Chen, Guang-Huei Lin, Pao-Ann Hsiung, Yu-Hen Hu, Hardware Software Co-Design of a Multimedia Soc Platform, page no 70 – 72

5. History of supercomputing, http://en.wikipedia.org/wiki/History_of_supercomputing searched on 23rd January, 2013.

6. http://www.ehow.com/facts_5191532_uses-supercomputer_.html , what are the uses of Supercomputer? , searched on 23rd January, 2013.

7. http://topsupercomputers-india.iisc.ernet.in/ , Top supercomputers India, searched on 27th January, 20138. http://www.ehow.com/info_12061579_purpose-supercomputer.html , what is the purpose of supercomputer,

searched on 26th January, 20139. http://en.wikipedia.org/wiki/Supercomputing_in_India , Supercomputing in India, searched on 23rd January,

2013.10. http://www.openloop.com/education/classes/sjsu_engr/engr_compOrg/spring2002/studentProjects/Andie_Hioki/Cray1withAdd.htm

, The Cray-1 Supercomputer, searched on 28th January, 2013.11. http://images.bit-tech.net/content_images/2010/07/cray-1-by-daryl-brach/cray-1-1-1280x1024.jpg, Images for

cray-1, searched on 6th February, 2013.12. http://www.computermuseum.org.uk/fixed_pages/cray_1_card.html, Cray 1 module, searched on 6th

February, 2013.13. http://www.cdac.in/index.aspx?id=pk_itn_spot844 , PARAM Yuva II, searched on 8th March 2013.14. http://www.cdac.in/index.aspx?id=pk_itn_spot851, PARAM Yuva II, searched on 8th March 2013.15. http://www.cdac.in/index.aspx?id=pk_itn_spot847, PARAM Yuva II, searched on 8th March 2013


http://en.wikipedia.org/wiki/History_of_supercomputing

http://www.ehow.com/facts_5191532_uses-supercomputer_.html

http://topsupercomputers-india.iisc.ernet.in/

http://www.ehow.com/info_12061579_purpose-supercomputer.html

http://en.wikipedia.org/wiki/Supercomputing_in_India

http://www.openloop.com/education/classes/sjsu_engr/engr_compOrg/spring2002/studentProjects/Andie_Hioki/Cray1withAdd.htm

http://images.bit-tech.net/content_images/2010/07/cray-1-by-daryl-brach/cray-1-1-1280x1024.jpg

http://www.computermuseum.org.uk/fixed_pages/cray_1_card.html

http://www.cdac.in/index.aspx?id=pk_itn_spot844




Asso. ProfessorDepartment of Computer science, Saurashtra University, Rajkot-360005, India

M- +91-9427236866, O- +91-281-2585232 [email protected] ,[email protected]


SUPERCOMPUTER TECHNOLOGIES IN MATHEMATICAL...

Documents

Transcript of SUPERCOMPUTER TECHNOLOGIES IN MATHEMATICAL...